Advertisement

Design, study quality and evidence of analgesic efficacy in studies of drugs in models of OA pain: a systematic review and a meta-analysis

  • A.K. Suokas
    Correspondence
    Address correspondence and reprint requests to: A.K. Suokas, Arthritis Research UK Pain Centre, Academic Rheumatology, Clinical Sciences Building, City Hospital, Nottingham NG5 1PB, UK. Tel: 44-(0)115-823-1092; Fax: 44-(0)115-823-1757.
    Affiliations
    Arthritis Research UK Pain Centre, University of Nottingham, Nottingham, UK

    School of Medicine, University of Nottingham, Nottingham, UK
    Search for articles by this author
  • D.R. Sagar
    Affiliations
    Arthritis Research UK Pain Centre, University of Nottingham, Nottingham, UK

    School of Life Sciences, University of Nottingham, Nottingham, UK
    Search for articles by this author
  • P.I. Mapp
    Affiliations
    Arthritis Research UK Pain Centre, University of Nottingham, Nottingham, UK

    School of Medicine, University of Nottingham, Nottingham, UK
    Search for articles by this author
  • V. Chapman
    Affiliations
    Arthritis Research UK Pain Centre, University of Nottingham, Nottingham, UK

    School of Life Sciences, University of Nottingham, Nottingham, UK
    Search for articles by this author
  • D.A. Walsh
    Affiliations
    Arthritis Research UK Pain Centre, University of Nottingham, Nottingham, UK

    School of Medicine, University of Nottingham, Nottingham, UK
    Search for articles by this author
Open ArchivePublished:July 04, 2014DOI:https://doi.org/10.1016/j.joca.2014.06.015

      Summary

      Objective

      Studies using animal models are important in drug development, but often poorly predict treatment results in man. We investigated factors that may impact on the magnitude of the analgesic treatment effect in animal models of osteoarthritis (OA) pain.

      Design

      Systematic review of studies that measured behavioural pain outcomes in small animal models of OA, and tested drugs which reduce OA pain in man. Standardised mean difference (SMD) and 95% confidence intervals (CIs) were calculated using random effects meta-analysis for selected models and drugs.

      Results

      Most studies used rat models (42/50) and chemical methods of OA induction (39/50). Analgesic treatment effect (SMD) was most commonly measured between drug- and vehicle treated rats with knee OA. Meta-analysis was carried out for 102 such comparisons from 26 studies. The pooled SMD was 1.36 (95% CI = 1.15–1.57). Non-steroidal anti-inflammatory drugs (NSAIDs) were associated with smaller SMDs than opioids (z = −3.25, P = 0.001). Grip strength gave larger SMDs than assessment of static weight bearing (z = −4.60, P < 0.001), mechanically-evoked pain (z = −3.83, P = 0.001) and movement-evoked pain (z = −5.23, P < 0.001), and SMDs for mechanically-evoked pain were larger than for movement-evoked pain (z = −2.78, P = 0.006). Studies that reported structural evaluation of OA phenotype were associated with smaller SMDs (z = −2.45, P = 0.014). Publication was significantly biased towards positive findings.

      Conclusion

      Attention to study-level moderators and publication bias may improve the ability of research using animal models to predict whether analgesic agents will reduce arthritis pain in man.

      Keywords

      Introduction

      Arthritis pain results from complex interactions between joint damage and inflammation, and abnormalities of pain processing
      • Gwilym S.E.
      • Pollard T.C.
      • Carr A.J.
      Understanding pain in osteoarthritis.
      • Mease P.J.
      • Hanna S.
      • Frakes E.P.
      • Altman R.D.
      Pain mechanisms in osteoarthritis: understanding the role of central pain and current approaches to its treatment.
      • Sofat N.
      • Ejindu V.
      • Kiely P.
      What makes osteoarthritis painful? The evidence for local and central pain processing.
      . Analgesic drugs commonly prescribed for arthritis target a variety of pain mechanisms, but may fail to provide adequate pain relief, or may be discontinued due to adverse events
      • Schnitzer T.J.
      Update on guidelines for the treatment of chronic musculoskeletal pain.
      . There is an urgent need to develop better analgesic drugs for people with arthritis.
      Preclinical testing is an essential part of drug development and licensing, and small animal models of osteoarthritis (OA)
      • Teeple E.
      • Jay G.D.
      • Elsaid K.A.
      • Fleming B.C.
      Animal models of osteoarthritis: challenges of model selection and analysis.
      • Gregory M.H.
      • Capito N.
      • Kuroki K.
      • Stoker A.M.
      • Cook J.L.
      • Sherman S.L.
      A review of translational animal models for knee osteoarthritis.
      are commonly used for developing treatments for arthritis pain. However, animal models of OA have displayed relatively low specificity for predicting analgesic agents that will reduce OA pain in man, and several drugs that have shown promise in preclinical studies have failed to fulfil that promise in clinical trials
      • Dworkin R.H.
      • Turk D.C.
      • Basch E.
      • Berger A.
      • Cleeland C.
      • Farrar J.T.
      • et al.
      Considerations for extrapolating evidence of acute and chronic pain analgesic efficacy.
      • Dworkin R.H.
      • Turk D.C.
      • Katz N.P.
      • Rowbotham M.C.
      • Peirce-Sandner S.
      • Cerny I.
      • et al.
      Evidence-based clinical trial design for chronic pain pharmacotherapy: a blueprint for ACTION.
      . Such failures are not exclusive to drug discovery for musculoskeletal pain as evidenced by systematic reviews
      • Perel P.
      • Roberts I.
      • Sena E.
      • Wheble P.
      • Briscoe C.
      • Sandercock P.
      • et al.
      Comparison of treatment effects between animal experiments and clinical trials: systematic review.
      • Sena E.
      • van der Worp H.B.
      • Howells D.
      • Macleod M.
      How can we improve the pre-clinical development of drugs for stroke?.
      • Sena E.S.
      • van der Worp H.B.
      • Bath P.M.W.
      • Howells D.W.
      • Macleod M.R.
      Publication bias in reports of animal stroke studies leads to major overstatement of efficacy.
      which have identified methodology and quality as major factors influencing study outcomes in preclinical studies. One of the key conclusions is that bias caused by design and quality factors, together with a tendency to publish only positive results, leads to an overestimation of drug efficacy. Low specificity in preclinical studies results in wasted resources on unsuccessful clinical trials, and unnecessary participant exposure to experimental drugs with uncertain toxicity. A better understanding of factors that contribute to overestimation of analgesic effects in preclinical studies should facilitate the development of novel effective treatments for arthritis pain in the future.
      We have undertaken a systematic review and meta-analysis to examine heterogeneity across preclinical studies of pharmacological interventions which reduce OA pain in man, and to investigate the impact of study methodology and quality on the magnitude of the observed analgesic treatment effect. A secondary aim was to identify suitable methodology to enable meta-analysis of preclinical studies without unnecessary loss of relevant data, despite the use of diverse pain-related behavioural tests across different stages of disease progression. This is, to our knowledge, the first meta-analysis of preclinical studies of the effects of interventions in models of OA pain.

      Methods

      Search strategy and selection of studies

      Four bibliographic databases (Medline 1948-, Embase 1980-, PubMed 1950- and Web of Science 1970-) were searched to identify studies published before July 2013. A systematic search strategy without language restrictions was developed in Medline (Appendix 1) and replicated in the other databases. The search strategy combined terms for OA, nociception, small animal models (i.e., the species), and models of OA induction (spontaneous or chemically induced or surgically induced). The review was limited to small animal models because investigation of disease mechanisms and initial screening of therapeutics typically involve small animals (mice, rats, guinea pigs, and rabbits) due to a faster disease progression, lower cost and ease of handling
      • Teeple E.
      • Jay G.D.
      • Elsaid K.A.
      • Fleming B.C.
      Animal models of osteoarthritis: challenges of model selection and analysis.
      • Gregory M.H.
      • Capito N.
      • Kuroki K.
      • Stoker A.M.
      • Cook J.L.
      • Sherman S.L.
      A review of translational animal models for knee osteoarthritis.
      . Reports were downloaded into Endnote X4 (Thomson Reuters, USA).
      Studies were eligible for inclusion if they reported the findings in a full paper published in a peer-reviewed academic journal and involved: (1) a prospective controlled experiment using mice, rats, Guinea pigs, or rabbits; (2) a treatment group with spontaneous or experimentally-induced OA; (3) a matched control group of animals with OA, sham-operated or vehicle-injected or naive animals; (4) testing the analgesic treatment effect of a drug from a class of pharmacological interventions recommended in the National Institute for Health and Care Excellence (NICE) guideline for OA pain
      • NICE
      The Care and Management of Osteoarthritis in Adults.
      : oral non-steroidal anti-inflammatory drugs (NSAIDs) including COX-2 inhibitors; paracetamol; opioids; intra-articular corticosteroid injections; topical NSAIDs; and topical capsaicin; (5) at least one of following pain-related behavioural test modalities (i.e., pain modalities): thermally-evoked pain; mechanically-evoked pain; movement-evoked pain; grip strength; and static weight bearing (Table I).
      Table IThe modalities, test methods and the outcome measures included in the review
      ModalityTest methodOutcome measures
      1. Thermally evoked painHot plate; a radiant heat sourceWithdrawal latency
      Acetone cooling testFlinching, shaking or toe-biting
      Cold plateWithdrawal latency
      2. Mechanically evoked painVon FreyWithdrawal threshold or latency
      PressureWithdrawal threshold; vocalisation; struggle
      3. Movement evoked painGait analysisDuty cycle; swing phase; swing speed; paw print intensity; stance; propulsion; limb idleness; speed
      Forced ambulation (rod)Use of the limb and presence of a limp
      Forced exerciseSlips
      Knee flexions and/or extensionsVocalisation; struggle
      4. Grip strengthGrip strengthThe maximum force displayed by an animal
      5. Static weight bearingIncapacitance testingHind paw weight distribution
      Abstracts were assessed by AKS and DRS, and full text was retrieved for all potentially relevant studies. Full papers were assessed to identify the species; OA models; treatment and control groups; drugs; and the pain modalities used in the studies. The meta-analysis was limited to a sub-group of studies to reduce heterogeneity among the experimental material. Firstly, the analysis was carried out on studies that compared drug- and vehicle-treated groups of rats with OA of the knee that is the most commonly investigated joint in OA models
      • Vincent T.L.
      • Williams R.O.
      • Maciewicz R.
      • Silman A.
      • Garside P.
      Mapping pathogenesis of arthritis through small animal models.
      . Secondly, only drugs tested in at least three studies using a systemic route of administration (oral, intravenous, subcutaneous, intramuscular) were included. Finally, studies were only included if they reported (either numerically or in a graph) the mean values and estimates of random variability for analgesic treatment effect at individual time points. Additional data on sample sizes were requested from the authors, if they were not reported fully in the papers
      • Boyce-Rustay J.M.
      • Simler G.H.
      • McGaraughty S.
      • Chu K.L.
      • Wensink E.J.
      • Vasudevan A.
      • et al.
      Characterization of fasudil in preclinical models of pain.
      • Kuduk S.D.
      • Di Marco C.N.
      • Bodmer-Narkevitch V.
      • Cook S.P.
      • Cato M.J.
      • Jovanovska A.
      • et al.
      Synthesis, structure-activity relationship, and pharmacological profile of analogs of the ASIC-3 inhibitor A-317567.
      • Pulichino A.M.
      • Rowland S.
      • Wu T.
      • Clark P.
      • Xu D.
      • Mathieu M.C.
      • et al.
      Prostacyclin antagonism reduces pain and inflammation in rodent models of hyperalgesia and chronic arthritis.
      • Pomonis J.D.
      • Boulet J.M.
      • Gottshall S.L.
      • Phillips S.
      • Sellers R.
      • Bunton T.
      • et al.
      Development and pharmacological characterization of a rat model of osteoarthritis pain.
      .

      Data extraction

      Data were extracted for study design including: experimental models of OA (spontaneous; chemical substance and dose; surgical method to induce OA); sample size per treatment and control group; strain, sex, age and weight of the animals; pain modalities; test methods; outcome measures; the drug and its dose, route, and date(s) of administration after the induction of OA; date(s) of outcome measurement after the induction of OA; and time to outcome measurement after the administration of the drug.
      If a study involved more than one comparison (for different drugs, doses, times points and/or outcome measures) data were extracted separately from each. Where data were presented only graphically, the mean values and estimates of random variability (standard error (SE) or standard deviation (SD)) for pain-related behavioural outcome measures were extracted using digital ruler software (Plot Digitizer 1.9, University of South Alabama, USA). Digital ruler software enabled the measurement of distances, areas and perimeters of figures on a computer screen
      • Hooijmans C.R.
      • de Vries R.B.M.
      • Rovers M.M.
      • Gooszen H.G.
      • Ritskes-Hoitinga M.
      The effects of probiotic supplementation on experimental acute pancreatitis: a systematic review and meta-analysis.
      • Hooijmans C.R.
      • Pasker-de Jong P.C.M.
      • de Vries R.B.M.
      • Ritskes-Hoitinga M.
      The effects of long-term Omega-3 fatty acid supplementation on cognition and Alzheimer's pathology in animal models of Alzheimer's disease: a systematic review and meta-analysis.
      . In line charts where the indicator lines for a specific time point were overlapping and difficult to read, SEs/SDs were extracted by reading the line with the highest estimate of variability.
      Data were extracted and coded by one reviewer (AKS), and a sub-set of key variables was validated by three co-investigators (DSR, PIM and VC).

      Quality assessment

      Quality of the studies was examined using a modified version of the criteria devised by CAMARADES
      • Sena E.
      • van der Worp H.B.
      • Howells D.
      • Macleod M.
      How can we improve the pre-clinical development of drugs for stroke?.
      . The checklist (Appendix 2) assessed potential bias judged by the quality of reporting and it included the following eight criteria: randomisation; allocation concealment; blinding; evidence of induced OA pain using the chosen behavioural outcome measure prior to pharmacological testing; evidence that the animals had structural OA (histological, macroscopic, microscopic or X-ray); sample size calculation; statement of conflict of interest; and statement of compliance with animal welfare regulations. Each item was scored as 1 if reported satisfactorily, and 0 if not (unclear risk of bias). The maximum score was 8.

      Data analysis

      Categorisation of study characteristics

      Each generic drug was categorised as oral NSAID, opioid, paracetamol, intra-articular corticosteroid injection, topical NSAIDs, or topical capsaicin. Administration was coded as acute (single dose) or chronic (at least one dose on two or more consecutive days), and the route as intraperitoneal, oral or subcutaneous. A chemical method to induce OA, intra-articular injection of mono-iodoacetate (MIA), was coded according to the dose (≤1, 1, 2 or ≥3 mg/kg). The timing of outcome measurement was categorised by weekly intervals to reflect the stages of OA disease progression: Week 1 (under 7 days); Week 2 (days 7–13); Week 3 (days 14–20); Week 4 (days 21–27); and Week 5 (Day 28 or later) post-induction of the OA. Methods used to measure pain outcomes were categorised into five modalities (Table I). The species were subcategorised according to sex and strain, and combined sample size for the treatment and the control group was coded by median-split as ≤14 or >14 animals. The age and weight of the animals were not categorised or analysed due to insufficient reporting in the studies.

      Meta-analysis of the analgesic treatment effect

      The animal experiments included in this systematic review tested classes of drugs that have been shown to be effective in the management of OA pain in man
      • NICE
      The Care and Management of Osteoarthritis in Adults.
      , and these experiments were used to investigate study-level moderators presuming primarily positive analgesic treatment effects. Factors associated with large effect sizes may indicate bias that results in overestimation of analgesic treatment effect, and therefore low specificity for predicting analgesic agents that will reduce OA pain in man.
      Meta-analysis using the random effect model
      • Borenstein M.
      • Hedges L.
      • Higgins J.
      • Rothstein H.
      Introduction to Meta Analysis.
      was carried out in StatsDirect 2.7.8 (StatsDirect, Altrincham, UK). Standardised mean difference (SMD) and 95% confidence intervals (CIs) were calculated for each comparison that assessed the analgesic treatment effect of a drug at a specific dose, using a pain-related behavioural outcome measure, between drug-and vehicle-treated rats with knee OA. The outcome measures included in the analysis were categorised according to the class of drug and positive analgesic treatment effect, i.e., reduction in pain behaviour, was measured as positive increasing SMDs.
      The SMDs and 95% CIs were pooled by disease progression (Weeks 1–5) and the five pain modalities (Table I) and presented in a forest plot. The total sample size and the numbers of studies and comparisons were counted for each week and modality. Non-overlap in CIs indicated statistically significant differences and further subgroup analyses were carried out using meta-regression. Publication bias was examined using a funnel plot and Egger's test. The Q test and I2 statistics were calculated to measure the degree of heterogeneity between studies. The Q test suggests a significant level of heterogeneity if statistically significant (P < 0.05) whereas the I2 value (0–100%) indicates the percentage of the heterogeneity across studies that is not due to chance.

      Meta-regression of study-level moderators

      Meta-regression
      • Borenstein M.
      • Hedges L.
      • Higgins J.
      • Rothstein H.
      Introduction to Meta Analysis.
      was used to explain heterogeneity across comparisons using SMD as the response variable and study design and quality as explanatory variables. All the independent variables were coded into dummy variables and included in the regression model as covariates, apart from the total quality score which was included as an interval variable. Regression parameters were tested individually for each group of dummies and the total quality score, and adjusted for the class of drug. Inverse variance for the meta-regression was calculated in MS Office Excel 2007. Meta-regression was carried out in SPSS21 (IBM, USA) using random effects metareg command which produces z-tests and P-values for testing the significance of regression coefficients.

      Avoidance of double-counting

      All the comparisons (as per each drug, dose, time point, time to outcome measurement, and test method) reported in the studies were treated as separate observations. This introduced a problem of double-counting as the studies often used one group of animals in more than one comparison (range 1–54 comparisons). Including only one observation from each study would have excluded relevant data from the meta-analysis, while pooling of data may have resulted in a loss of information about the variation in the analgesic treatment effect. Therefore we pragmatically limited the number of observations selected for the analysis.
      For each drug, both the acute and the chronic administration were included from each week (weeks 1–5) for each behavioural pain test. Where analgesic treatment effect was measured at different doses of the drug, the median dose was chosen (celecoxib 30 mg/kg, diclofenac 30 mg/kg, indomethacin 2 mg/kg, naproxen 10 mg/kg, rofecoxib 10 mg/kg, morphine 3 mg/kg, and tramadol 100 mg/kg). The median dose was chosen assuming that it would be closest to the optimal dose and most relevant for the accurate measurement of analgesic treatment effect.
      Only one time point was included from each week (day 3, 7, 14, 21, 28 or the closest day after), and the median outcome measurement (1 h time) was chosen over other times. Decision to differentiate between weekly time points was made assuming that pain outcomes depend on disease progression
      • Gregory M.H.
      • Capito N.
      • Kuroki K.
      • Stoker A.M.
      • Cook J.L.
      • Sherman S.L.
      A review of translational animal models for knee osteoarthritis.
      , and that the analgesic treatment effect may change over time.
      Finally, only one set of results was included from repeat tests at increasing intensities. This concerned one study
      • Vonsy J.L.
      • Ghandehari J.
      • Dickenson A.H.
      Differential analgesic effects of morphine and gabapentin on behavioural measures of pain and disability in a model of osteoarthritis pain in rats.
      from which only data using the middle filament weight (6 g) of the three reported von Frey tests was included.
      This process decreased the number of observations from 246 to 102. The extent of residual double-counting was estimated assuming that each drug at a specific dose was tested in a separate cohort of animals, but that repeat tests were carried out for the following comparisons: (1) testing one group of animals at different weekly time points; (2) testing both the acute and chronic administration of a drug with one group of animals; (3) subjecting one group of animals to more than one behavioural pain test; and (4) including one vehicle-treated control group of animals in more than one comparison.

      Results

      Study selection

      A flow diagram in Fig. 1 presents the study selection process. Characteristics of the fifty studies that met the inclusion criteria are presented in Table II.
      Table IISummary of study characteristics in the 50 studies that met the inclusion criteria
      Study characteristicSub-groupsNumber of studies
      SpeciesRat42
      Mouse3
      Rabbit3
      Guinea pig2
      JointKnee45
      Facet2
      Shoulder2
      Ankle1
      OA modelChemical (all MIA
      MIA = Intra-articular injection of mono-iodoacetate.
      )
      39
      Surgical, of which11
      •  meniscal transection
       5
      •  anterior cruciate ligament section
       3
      •  medial meniscus destabilisation
       2
      •  Hulth model
        Hulth model: anterior and posterior cruciate ligaments, medial collateral ligament and meniscal transaction.
       1
      Spontaneous0
      Control group
      One study could involve more than one control group, drug and pain modality.
      OA48
       of which vehicle-treated OA
      Vehicle-treated OA = animals with OA that receive the vehicle for the delivery of the drug (such as saline) instead of the drug.
       46
      Sham-induced
      Sham-induced = Healthy control animals that receive sham surgery/injections so that the study can control for the effect of anaesthesia and the invasive procedure involved in the induction of OA. Sham procedures take place prior to the pharmacological study.
      15
      Naïve
      Naïve = Healthy control animals that do not undergo any sham procedures prior to the pharmacological study.
      12
      Drugs tested
      One study could involve more than one control group, drug and pain modality.
      NSAID31
      Opioid16
      Paracetamol4
      Topical NSAID2
      Topical capsaicin0
      Intra-articular corticosteroid injection0
      Pain modality
      One study could involve more than one control group, drug and pain modality.
      Static weight bearing28
      Mechanically evoked pain15
      Grip strength7
      Movement evoked pain6
      Thermally evoked pain3
      One study could involve more than one control group, drug and pain modality.
      MIA = Intra-articular injection of mono-iodoacetate.
      Hulth model: anterior and posterior cruciate ligaments, medial collateral ligament and meniscal transaction.
      § Vehicle-treated OA = animals with OA that receive the vehicle for the delivery of the drug (such as saline) instead of the drug.
      || Sham-induced = Healthy control animals that receive sham surgery/injections so that the study can control for the effect of anaesthesia and the invasive procedure involved in the induction of OA. Sham procedures take place prior to the pharmacological study.
      Naïve = Healthy control animals that do not undergo any sham procedures prior to the pharmacological study.
      The majority (42/50) of the studies used rat model of OA and only a few (3, 3 and 2, respectively, of 50) used mice, rabbits or guinea pigs, and knee was the most commonly investigated joint (45/50). Thirty-nine studies used a chemical method of induction which was intra-articular injection of MIA. Eleven studies used surgical method of induction which included meniscal transection (5/11), anterior cruciate ligament section (3/11), destabilisation of the medial meniscus (2/11) and the Hulth model (1/11). The drugs tested in the studies included NSAIDs (31/50), opioids (16/50), paracetamol (4/50) or topical NSAID (2/50). The most common pain modalities reported were static weight bearing (28/50) and mechanically-evoked pain (15/50), and smaller number of studies examined grip strength (7/50), movement-evoked pain (6/50) and thermally-evoked pain (3/50). Individual studies could involve more than one control group, drug, and pain modality (i.e., the total for each item may exceed 50).
      The majority (46/50) of the studies compared the analgesic treatment effect of drugs with vehicle-treatment in animals with OA, and meta-analysis was carried out for 102 such comparisons from 26 studies that investigated a rat model of knee OA. The following drugs were tested in at least three studies and satisfied our criteria for meta-analysis (Table III): celecoxib (10/26), diclofenac (7/26), morphine (7/26), rofecoxib (4/26), indomethacin (3/26), naproxen (3/26) and tramadol (3/26). The included drugs were either NSAIDs or opioids.
      Table IIISummary of study characteristics in the 26 studies and 102 comparisons included in the meta-analysis. Each comparison compared a drug-treated group and a vehicle-treated group of rats with knee OA, and assessed the analgesic treatment effect of a drug at a specific dose
      CharacteristicSub-groupsNumber of studiesNumber of comparisons
      Animal model of OAMIA
      MIA = Intra-articular injection of mono-iodoacetate.
      2294
       1 mg729
       2 mg524
       ≥3 mg1041
      Meniscal transection37
      Anterior cruciate ligament section11
      Rat strainSprague Dawley1652
      Wistar840
      Lewis26
      Sex of the ratsMale26102
      Female00
      Drugs tested
      One study could involve more than one drug, type of administration and behavioural pain test.
      ,
      The following drugs were tested in fewer than three papers and thus excluded: etoricoxib, felbinac, ibuprofen, ketoprofen, loxoprofen, meloxicam, nimesulide and paracetamol.
      NSAID2374Dose mg/kg
       Celecoxib103810–100
       Diclofenac71730
       Rofecoxib4510
       Indomethacin362–3
       Naproxen3810
      Opioid928
       Morphine7221–6
       Tramadol3630–100
      Route of administration
      One study could involve more than one drug, type of administration and behavioural pain test.
      Oral1864
      Subcutaneous734
      Intraperitoneal44
      Length of administration
      Length of administration: acute = one time dose, chronic = at least one dose on two or more consecutive days.
      ,
      One study could involve more than one drug, type of administration and behavioural pain test.
      Acute2267
      Chronic835
      Pain modality
      One study could involve more than one drug, type of administration and behavioural pain test.
      Static weight bearing1546
      Mechanically evoked pain631
      Pressure314
      von Frey517
      Movement evoked pain318
       Gait analysis214
       Forced ambulation12
       Knee flexions and/or extensions12
      Grip strength55
      Thermally evoked pain (acetone cooling test)12
      Total26 studies102 comparisons
      Numbers for the main categories are printed in bold, while the numbers for the sub-categories are in plain text.
      MIA = Intra-articular injection of mono-iodoacetate.
      Length of administration: acute = one time dose, chronic = at least one dose on two or more consecutive days.
      One study could involve more than one drug, type of administration and behavioural pain test.
      § The following drugs were tested in fewer than three papers and thus excluded: etoricoxib, felbinac, ibuprofen, ketoprofen, loxoprofen, meloxicam, nimesulide and paracetamol.

      Characteristics of studies included in the meta-analysis

      The characteristics of the 26 studies and 102 comparisons included in the meta-analysis are summarised in Table III, and the studies are listed in Table IV. OA was induced using MIA in 94 of the 102 comparisons, and only eight comparisons used a surgical method of induction. The dose of MIA used to induce OA was 1 mg (29/102), 2 mg (24/102) or ≥3 mg (41/102). Most of the comparisons used Sprague Dawley (52/102) or Wistar (40/102) rats. The pain modalities were static weight bearing (46/102), mechanically-evoked pain (31/102), movement-evoked pain (18/102), grip strength (5/102), and thermally-evoked pain (2/102). The majority of the comparisons tested NSAIDs (74/102). The drugs and doses tested are presented in Table III.
      Table IVList of the studies and comparisons included in the meta-analysis and meta-regression
      StudyNo. of comparisonsSample size per group Experimental/ControlModel
      MIA = monosodium iodoacetate, ACLT = anterior cruciate ligament transection, MNX = meniscal transection.
      MIA mg/kgRat strain
      Rat strain: L = Lewis, SD = Sprague Dawley, W = Wistar.
      Generic drugDose mg/kgRoute
      Route: IP = intraperitoneal, O = oral, SC = subcutaneous.
      Date(s) givenTest date(s)Behavioural pain test(s)
      Ahn 2012
      • Ahn K.
      • Smith S.E.
      • Liimatta M.B.
      • Beidler D.
      • Sadagopan N.
      • Dudley D.T.
      • et al.
      Mechanistic and pharmacological characterization of PF-04457845: a highly potent and selective fatty acid amide hydrolase inhibitor that reduces inflammatory and noninflammatory pain.
      11010MIA2SDCelecoxib30O1414Joint compression
      Ahn 2012
      • Ahn K.
      • Smith S.E.
      • Liimatta M.B.
      • Beidler D.
      • Sadagopan N.
      • Dudley D.T.
      • et al.
      Mechanistic and pharmacological characterization of PF-04457845: a highly potent and selective fatty acid amide hydrolase inhibitor that reduces inflammatory and noninflammatory pain.
      11010MIA2SDCelecoxib30O14–1616Joint compression
      Ashraf 2011
      • Ashraf S.
      • Mapp P.I.
      • Walsh D.A.
      Contributions of angiogenesis to inflammation, joint damage, and pain in a rat model of osteoarthritis.
      488MNXN/ALIndomethacin2O11–3512, 14, 26, 29Weight bearing
      Beyreuther 2007
      • Beyreuther B.
      • Callizot N.
      • Stohr T.
      Antinociceptive efficacy of lacosamide in the monosodium iodoacetate rat model for osteoarthritis pain.
      61212MIA3WDiclofenac30SC3, 7, 143, 7, 14Paw pressure; von Frey
      Beyreuther 2007
      • Beyreuther B.
      • Callizot N.
      • Stohr T.
      Antinociceptive efficacy of lacosamide in the monosodium iodoacetate rat model for osteoarthritis pain.
      61212MIA3WMorphine3SC3, 7, 143, 7, 14Paw pressure; von Frey
      Bley 2006
      • Bley K.R.
      • Bhattacharya A.
      • Daniels D.V.
      • Gever J.
      • Jahangir A.
      • O'Yang C.
      • et al.
      RO1138452 and RO3244794: characterization of structurally distinct, potent and selective IP (prostacyclin) receptor antagonists.
      11010MIA1WRofecoxib10O1414Weight bearing
      Bove 2003
      • Bove S.E.
      • Calcaterra S.L.
      • Brooker R.M.
      • Huber C.M.
      • Guzman R.E.
      • Juneau P.L.
      • et al.
      Weight bearing as a measure of disease progression and efficacy of anti-inflammatory compounds in a model of monosodium iodoacetate-induced osteoarthritis.
      148MIA1WNaproxen10O1414Weight bearing
      Bove 2003
      • Bove S.E.
      • Calcaterra S.L.
      • Brooker R.M.
      • Huber C.M.
      • Guzman R.E.
      • Juneau P.L.
      • et al.
      Weight bearing as a measure of disease progression and efficacy of anti-inflammatory compounds in a model of monosodium iodoacetate-induced osteoarthritis.
      188MIA1WRofecoxib10O1414Weight bearing
      Bove 2006
      • Bove S.E.
      • Laemont K.D.
      • Brooker R.M.
      • Osborn M.N.
      • Sanchez B.M.
      • Guzman R.E.
      • et al.
      Surgically induced osteoarthritis in the rat results in the development of both osteoarthritis-like joint pain and secondary hyperalgesia.
      286–8MNXN/ALRofecoxib10O2121Weight bearing, von Frey
      Boyce-Rustay 2010
      • Boyce-Rustay J.M.
      • Simler G.H.
      • McGaraughty S.
      • Chu K.L.
      • Wensink E.J.
      • Vasudevan A.
      • et al.
      Characterization of fasudil in preclinical models of pain.
      11212MIA3SDCelecoxib30IP20 or later20 or laterGrip force
      Castro 2006
      • Castro R.R.
      • Cunha F.Q.
      • Silva F.S.
      • Rocha F.A.C.
      A quantitative approach to measure joint pain in experimental osteoarthritis – evidence of a role for nitric oxide.
      166ACLTN/AWIndomethacin2SC4–74–7Weight bearing
      Combe 2004
      • Combe R.
      • Bramwell S.
      • Field M.J.
      The monosodium iodoacetate model of osteoarthritis: a model of chronic nociceptive pain in rats?.
      11010MIA2SDMorphine3SC14 or later14 or laterWeight bearing
      Combe 2004
      • Combe R.
      • Bramwell S.
      • Field M.J.
      The monosodium iodoacetate model of osteoarthritis: a model of chronic nociceptive pain in rats?.
      11010MIA2SDTramadol10SC14 or later14 or laterWeight bearing
      Cowart 2012
      • Cowart M.
      • Hsieh G.
      • Black L.A.
      • Zhan C.
      • Gomez E.J.
      • Pai M.
      • et al.
      Pharmacological characterization of A-960656, a histamine H(3) receptor antagonist with efficacy in animal models of osteoarthritis and neuropathic pain.
      11212MIA3SDDiclofenac30O2121Grip force
      Ferland 2011
      • Ferland C.E.
      • Laverty S.
      • Beaudry F.
      • Vachon P.
      Gait analysis and pain response of two rodent models of osteoarthritis.
      1666MIA3SDCelecoxib30O1–283, 8, 15, 22Gait duty cycle/swing phase/swing speed, von Frey
      Fernihough 2004
      • Fernihough J.
      • Gentry C.
      • Malcangio M.
      • Fox A.
      • Rediske J.
      • Pellas T.
      • et al.
      Pain related behaviour in two models of osteoarthritis in the rat knee.
      566MIA2WDiclofenac30SC3, 14, 283, 14, 28Paw pressure, von Frey only on days 14 and 28
      Fernihough 2004
      • Fernihough J.
      • Gentry C.
      • Malcangio M.
      • Fox A.
      • Rediske J.
      • Pellas T.
      • et al.
      Pain related behaviour in two models of osteoarthritis in the rat knee.
      566MIA2WMorphine6SC3, 14, 283, 14, 28Paw pressure, von Frey only on days 14 and 28
      Ferreira-Gomez 2012
      • Ferreira-Gomes J.
      • Adaes S.
      • Mendonca M.
      • Castro-Lopes J.M.
      Analgesic effects of lidocaine, morphine and diclofenac on movement-induced nociception, as assessed by the Knee-Bend and CatWalk tests in a rat model of osteoarthritis.
      254MIA2WDiclofenac30O33Gait, knee bend test
      Ferreira-Gomez 2012
      • Ferreira-Gomes J.
      • Adaes S.
      • Mendonca M.
      • Castro-Lopes J.M.
      Analgesic effects of lidocaine, morphine and diclofenac on movement-induced nociception, as assessed by the Knee-Bend and CatWalk tests in a rat model of osteoarthritis.
      254MIA2WMorphine6SC33Gait, knee bend test
      Fitzpatrick 2011
      • Fitzpatrick L.R.
      • Green C.
      • Maines L.W.
      • Smith C.D.
      Experimental osteoarthritis in rats is attenuated by ABC294640, a selective inhibitor of sphingosine kinase-2.
      449MIA3WTramadol100O5, 7, 14, 285, 7, 14, 28Weight bearing
      Hsieh 2010A
      • Hsieh G.C.
      • Chandran P.
      • Salyers A.K.
      • Pai M.
      • Zhu C.Z.
      • Wensink E.J.
      • et al.
      H4 receptor antagonism exhibits anti-nociceptive effects in inflammatory and neuropathic pain models in rats.
      166MIA3SDCelecoxib30IP2020Grip force
      Hsieh 2010B
      • Hsieh G.C.
      • Honore P.
      • Pai M.
      • Wensink E.J.
      • Chandran P.
      • Salyers A.K.
      • et al.
      Antinociceptive effects of histamine H3 receptor antagonist in the preclinical models of pain in rats and the involvement of central noradrenergic systems.
      11212MIA3SDCelecoxib30IP2020Grip force
      Ivanavicius 2007
      • Ivanavicius S.P.
      • Ball A.D.
      • Heapy C.G.
      • Westwood F.R.
      • Murray F.
      • Read S.J.
      Structural pathology in a rodent model of osteoarthritis is associated with neuropathic pain: increased expression of ATF-3 and pharmacological characterisation.
      366MIA1WCelecoxib10O14, 21, 2814, 21, 28Weight bearing
      Ivanavicius 2007
      • Ivanavicius S.P.
      • Ball A.D.
      • Heapy C.G.
      • Westwood F.R.
      • Murray F.
      • Read S.J.
      Structural pathology in a rodent model of osteoarthritis is associated with neuropathic pain: increased expression of ATF-3 and pharmacological characterisation.
      366MIA1WNaproxen10O14, 21, 2814, 21, 28Weight bearing
      Kang 2010
      • Kang M.
      • Jung I.
      • Hur J.
      • Kim S.H.
      • Lee J.H.
      • Kang J.Y.
      • et al.
      The analgesic and anti-inflammatory effect of WIN-34B, a new herbal formula for osteoarthritis composed of Lonicera japonica Thunb and Anemarrhena asphodeloides BUNGE in vivo.
      31010MIA3SDCelecoxib100O1–144, 7, 14Weight bearing
      Okun 2012
      • Okun A.
      • Liu P.
      • Davis P.
      • Ren J.
      • Remeniuk B.
      • Brion T.
      • et al.
      Afferent drive elicits ongoing pain in a model of advanced osteoarthritis.
      186MIA4.8SDDiclofenac30O1414Weight bearing
      Plummer 2013
      • Plummer M.S.
      • Cornicelli J.
      • Roark H.
      • Skalitzky D.J.
      • Stankovic C.J.
      • Bove S.
      • et al.
      Discovery of potent selective bioavailable phosphodiesterase 2 (PDE2) inhibitors active in an osteoarthritis pain model. Part II: optimization studies and demonstration of in vivo efficacy.
      166MNXN/ASDRofecoxib10O2828Weight bearing
      Pomonis 2005
      • Pomonis J.D.
      • Boulet J.M.
      • Gottshall S.L.
      • Phillips S.
      • Sellers R.
      • Bunton T.
      • et al.
      Development and pharmacological characterization of a rat model of osteoarthritis pain.
      11010MIA1SDCelecoxib30O2222Weight bearing
      Pomonis 2005
      • Pomonis J.D.
      • Boulet J.M.
      • Gottshall S.L.
      • Phillips S.
      • Sellers R.
      • Bunton T.
      • et al.
      Development and pharmacological characterization of a rat model of osteoarthritis pain.
      11010MIA1SDCelecoxib30O10–2222Weight bearing
      Pomonis 2005
      • Pomonis J.D.
      • Boulet J.M.
      • Gottshall S.L.
      • Phillips S.
      • Sellers R.
      • Bunton T.
      • et al.
      Development and pharmacological characterization of a rat model of osteoarthritis pain.
      11010MIA1SDIndomethacin3O2222Weight bearing
      Pomonis 2005
      • Pomonis J.D.
      • Boulet J.M.
      • Gottshall S.L.
      • Phillips S.
      • Sellers R.
      • Bunton T.
      • et al.
      Development and pharmacological characterization of a rat model of osteoarthritis pain.
      11010MIA1SDMorphine3SC2121Weight bearing
      Pulichino 2006
      • Pulichino A.M.
      • Rowland S.
      • Wu T.
      • Clark P.
      • Xu D.
      • Mathieu M.C.
      • et al.
      Prostacyclin antagonism reduces pain and inflammation in rodent models of hyperalgesia and chronic arthritis.
      11929MIA1SDDiclofenac10O77Weight bearing
      Rashid 2013
      • Rashid M.H.
      • Theberge Y.
      • Elmes S.J.
      • Perkins M.N.
      • McIntosh F.
      Pharmacological validation of early and late phase of rat mono-iodoacetate model using the Tekscan system.
      488MIA1SDCelecoxib38OWeeks 1, 2, 3, 4Weeks 1, 2, 3, 4Weight bearing
      Rashid 2013
      • Rashid M.H.
      • Theberge Y.
      • Elmes S.J.
      • Perkins M.N.
      • McIntosh F.
      Pharmacological validation of early and late phase of rat mono-iodoacetate model using the Tekscan system.
      488MIA1SDCelecoxib38OWeeks 1, 2, 3, 4Weeks 1, 2, 3, 4Weight bearing
      Rashid 2013
      • Rashid M.H.
      • Theberge Y.
      • Elmes S.J.
      • Perkins M.N.
      • McIntosh F.
      Pharmacological validation of early and late phase of rat mono-iodoacetate model using the Tekscan system.
      288MIA1SDNaproxen50OWeeks 1, 3Weeks 1, 3Weight bearing
      Rashid 2013
      • Rashid M.H.
      • Theberge Y.
      • Elmes S.J.
      • Perkins M.N.
      • McIntosh F.
      Pharmacological validation of early and late phase of rat mono-iodoacetate model using the Tekscan system.
      288MIA1SDNaproxen50OWeeks 1, 3Weeks 1, 3Weight bearing
      Rashid 2013
      • Rashid M.H.
      • Theberge Y.
      • Elmes S.J.
      • Perkins M.N.
      • McIntosh F.
      Pharmacological validation of early and late phase of rat mono-iodoacetate model using the Tekscan system.
      188MIA1SDMorphine1OWeeks 1, 2, 3, 4Weeks 1, 2, 3, 4Weight bearing
      Vonsy 2009
      • Vonsy J.L.
      • Ghandehari J.
      • Dickenson A.H.
      Differential analgesic effects of morphine and gabapentin on behavioural measures of pain and disability in a model of osteoarthritis pain in rats.
      377MIA2SDMorphine3SC1414Cooling hypersensitivity, forced ambulation, von Frey
      Vonsy 2009
      • Vonsy J.L.
      • Ghandehari J.
      • Dickenson A.H.
      Differential analgesic effects of morphine and gabapentin on behavioural measures of pain and disability in a model of osteoarthritis pain in rats.
      377MIA2SDMorphine3SC14–1818Cooling hypersensitivity, forced ambulation, von Frey
      Yao 2008
      • Yao B.B.
      • Hsieh G.C.
      • Frost J.M.
      • Fan Y.
      • Garrison T.R.
      • Daza A.V.
      • et al.
      In vitro and in vivo characterization of A-796260: a selective cannabinoid CB2 receptor agonist exhibiting analgesic activity in rodent pain models.
      166MIA3SDCelecoxib38IP2020Grip force
      Yoshimi 2010
      • Yoshimi E.
      • Kumakura F.
      • Hatori C.
      • Hamachi E.
      • Iwashita A.
      • Ishii N.
      • et al.
      Antinociceptive effects of AS1892802, a novel Rho kinase inhibitor, in rat models of inflammatory and noninflammatory arthritis.
      188MIA1SDDiclofenac30O21 or later21 or laterWeight bearing
      Yoshimi 2010
      • Yoshimi E.
      • Kumakura F.
      • Hatori C.
      • Hamachi E.
      • Iwashita A.
      • Ishii N.
      • et al.
      Antinociceptive effects of AS1892802, a novel Rho kinase inhibitor, in rat models of inflammatory and noninflammatory arthritis.
      188MIA1SDTramadol10O21 or later21 or laterWeight bearing
      Total comparisons102
      The highlighted comparisons tested chronic (at least one dose on two or more consecutive days) administration of the analgesic drug.
      MIA = monosodium iodoacetate, ACLT = anterior cruciate ligament transection, MNX = meniscal transection.
      Rat strain: L = Lewis, SD = Sprague Dawley, W = Wistar.
      Route: IP = intraperitoneal, O = oral, SC = subcutaneous.

      Meta-analysis of the analgesic treatment effect

      The analgesic treatment effect for all 102 comparisons (pooled SMD; Appendix 3) was 1.36 (95%CI = 1.15–1.57) and the total sample size was 1628. The distribution of SMDs was positively skewed and included some large values for the observed treatment effect. The treatment effect for NSAIDs was significantly lower (pooled SMD 1.16; 95%CI = 0.91–1.40) than for opioids (pooled SMD 1.90; 95%CI = 1.56–2.23). The pooled SMDs for each drug class were similar across time (Appendix 4). Pooled analgesic treatment effects of NSAIDs and opioids by pain modality are presented in Fig. 2 (forest plots in Appendix 5). Thermally-evoked pain with only two observations from one study was excluded from the analysis. The pooled SMD (95%CI) was highest for grip strength at 3.96 (2.38–5.55) and lowest for movement-evoked pain at 0.31 (−0.08 to 0.71). Other pain modalities displayed intermediate SMDs.
      Figure thumbnail gr2
      Fig. 2Pooled SMDs and 95% CIs for the analgesic treatment effect of NSAIDs and opioids, by pain modality. Positive increasing SMDs indicate analgesic efficacy. Thermally-evoked pain with only two observations from one study was excluded from the analysis. One study may involve the testing of more than one pain modality and/or drug. Forest plots for each subgroup are presented in .
      Across the 102 comparisons, the funnel plot (Fig. 3) and Egger's test (P < 0.001) showed publication bias due to a small number of studies reporting negative analgesic treatment effects. Both the Q test (Q = 340.55, P < 0.001) and I2 value (70%, 95% CI = 64–75%) suggested a significant level of heterogeneity, which reflects the diversity of study design among the 102 comparisons.
      Figure thumbnail gr3
      Fig. 3Funnel plot for the magnitude of the analgesic treatment effect (n = 102). Test for publication bias: Egger's test = 5.04 (95% CI = 3.72–6.36) P < 0.0001.

      Meta-regression of study-level moderators

      Overall, the class of drug, pain modality, route of administration and study quality emerged as significant explanatory variables in the meta-regression (Table V, Appendix 6). As found in the meta-analysis, NSAIDs were associated with smaller analgesic treatment effects (SMDs) than opioids (z = −3.25, P = 0.001). Grip strength gave larger SMDs than assessment of static weight bearing (z = −4.60, P < 0.001), mechanically-evoked pain (z = −3.83, P = 0.001) and movement-evoked pain (z = −5.23, P < 0.001). Further, the SMDs for mechanically-evoked pain were larger than for movement-evoked pain (z = −2.78, P = 0.006). Intraperitoneal route of administration was associated with higher SMDs than oral and subcutaneous (z = −3.94, P < 0.001 and z = −3.44, P < 0.001). However, intraperitoneal route was only used in four studies that all tested grip strength with 30–38 mg/kg of celecoxib and gave large SMDs (range 1.73–7.78). Due to the small numbers it is not possible to investigate whether this route improved the SMDs for grip strength.
      Table VStudy-level moderators of the magnitude of the analgesic treatment effect. Full results are presented in Appendix 6
      Overall model (n = 102)BZP
      Individual variables/groups of dummies
      Class of drug
      NSAID vs opioid−0.80−3.250.001
      Pain modality
      Controlled for class of drugs.
      ,
      Thermally-evoked pain with only two observations from one study was excluded from the analysis, n = 100.
      Weight bearing vs grip strength−2.66−4.60<0.001
      Mechanically-evoked pain vs grip strength−2.29−3.83<0.001
      Movement-evoked pain vs grip strength−3.22−5.23<0.001
      Movement-evoked pain vs mechanically-evoked pain−0.92−2.780.006
      Route of administration
      Controlled for class of drugs.
      Oral vs intraperitoneal−2.55−3.94<0.001
      Subcutaneous vs intraperitoneal−2.36−3.44<0.001
      Study quality sub-components
      Controlled for class of drugs.
      Evidence of OA−0.61−2.450.014
      Overall study quality
      Controlled for class of drugs.
      Total quality score−0.34−3.64<0.001
      Controlled for class of drugs.
      Thermally-evoked pain with only two observations from one study was excluded from the analysis, n = 100.
      Incomplete reporting limited the assessment of study quality and its impact on the magnitude of the SMDs. None of the studies reported sample size calculation or allocation concealment, and less than half (12/26) reported randomisation and/or blinding (Table VI). The quality scores varied between 0 and 5 and the average score per study was 3.7 out of 8. Studies that achieved a higher overall score and studies that presented evidence that the animals had structural OA were associated with smaller SMDs (z = −3.64, P < 0.001 and z = −2.45, P = 0.014 respectively). No association was found between the SMDs and the other design features or quality indicators (results not presented).
      Table VINumber and proportion of publications reporting individual components of the study quality checklist for the 26 studies included in the meta-analysis, and for the total 50 studies. The checklist is presented in Appendix 2
      Criteria26 studies in the meta-analysisTotal of 50 studies
      n%n%
      Randomisation12461938
      Allocation concealment0000
      Blinding12461836
      Evidence of induced OA pain10392142
      Evidence that the animals had OA13502550
      Sample size calculation0000
      Statement regarding potential conflict of interest3121224
      Statement of compliance with animal welfare regulations25964692

      Discussion

      This review has identified that study-level moderators and publication bias may result in an overestimation of analgesic treatment effect in animal models of OA. Bias towards positive results increases the likelihood of identifying promising therapies that fail in clinical trials. In preclinical studies that model pain, such bias can be defined as low specificity to predict analgesic agents that will reduce pain in man. We have identified three factors – choice of design, limited study quality and publication bias – that may contribute to low specificity and thus failure of clinical prediction.

      Choice of design

      Pain modality and drug administration route emerged significant in the meta-regression. Variation in SMDs across the five modalities may indicate differences in measurement properties between the pain-related behaviours. Weight bearing evaluates pain on loading the joint at rest, in contrast to pain behaviour observed on joint movement. Thermally- and mechanically-evoked pain tests are used to measure hyperalgesia, an increased sensitivity to painful stimuli, or allodynia, pain resulting from a stimulus which would not normally provoke pain. Grip strength measures muscle strength which may be inhibited by pain. The various behavioural measures could therefore have different predictive value for analgesic efficacy in different patient subgroups, or, within an individual, for different aspects of their pain experience.
      The pooled SMD for grip strength was significantly higher than for the other modalities, suggesting that grip strength may be highly sensitive for measuring OA pain in rats. Although there is potential confounding by intraperitoneal route of drug administration, this is unlikely to explain the high effect sizes achieved using grip strength tests. Intraperitoneal injection enables faster absorption than oral administration
      • Boersen N.
      • Lee T.
      • Hui H.
      Development of preclinical formulations for toxicology studies.
      , but the proportion of the drug that enters the circulation is lower than by the subcutaneous route
      • Hedenqvist P.
      • Hellebrekers L.J.
      Laboratory animal analgesia, anesthesia, and euthanasia.
      • Turner P.V.
      • Brabb T.
      • Pekow C.
      • Vasbinder M.A.
      Administration of substances to laboratory animals: routes of administration and factors to consider.
      . Grip strength was developed to assess the integrity of the neuromuscular system in toxicology studies, but more recently has been used in animal models of pain
      • Kehl L.J.
      • Trempe T.M.
      • Hargreaves K.M.
      A new animal model for assessing mechanisms and management of muscle hyperalgesia.
      • Wacnik P.W.
      • Kehl L.J.
      • Trempe T.M.
      • Ramnaraine M.L.
      • Beitz A.J.
      • Wilcox G.L.
      Tumor implantation in mouse humerus evokes movement-related hyperalgesia exceeding that evoked by intramuscular carrageenan.
      • Chandran P.
      • Pai M.
      • Blomme E.A.
      • Hsieh G.C.
      • Decker M.W.
      • Honore P.
      Pharmacological modulation of movement-evoked pain in a rat model of osteoarthritis.
      . Our analysis suggests that grip strength deserves further attention as a sensitive pain behavioural measure in rats, with potential to reduce the numbers of animals required to demonstrate efficacy in analgesic studies.

      Publication bias and study quality

      The distribution of SMDs was positively skewed and included some large values for positive treatment effects, which suggests under-reporting of less positive results. Publication bias may be generated by selective reporting by authors, or journal editorial policy on concise reporting. Publication of well conducted but negative studies is important in order to provide a balanced picture of analgesic efficacy within the scientific literature.
      Lower study quality was associated with larger SMDs and potential overestimation of the analgesic treatment effect. Lack of reported evidence that OA structural change was successfully induced in the model was strongly associated with larger effect sizes. Incomplete phenotyping of animals prior to pharmacological interventions may result in confounding of analgesic effects by non-OA pathology. Evidence of structural OA is a pragmatic and relevant measure of study quality that can be used alongside the traditional checklist items. Indeed, traditional quality items such as blinding and randomisation were rarely reported and failed to explain differences in the analgesic treatment effect.

      Caveats of the study

      Systematic review and meta-analysis are now well recognised methodologies permitting robust conclusions in clinical sciences, but raise specific issues in preclinical studies. Repeat observations are typically reported in each preclinical study, and a single control group is often used in more than one comparison, thereby reducing animal usage. However, inclusion of multiple observations per group and per study may introduce bias during meta-analysis. A further limitation of our work is that preclinical studies frequently did not report data in numerical format, necessitating the use of digital ruler software that may result in measurement errors. We concur with the Animal Research: Reporting In Vivo Experiments (ARRIVE) guidelines
      • Kilkenny C.
      • Browne W.J.
      • Cuthill I.C.
      • Emerson M.
      • Altman D.G.
      Improving bioscience research reporting: the ARRIVE guidelines for reporting animal research.
      which recommend improved reporting standards in preclinical research.
      It is worth noting that we did not assess the predictive value of individual studies, and that a large analgesic treatment effect does not mean failure of prediction. Rather, we used a set of observations that tested NSAIDs and opioids to identify study-level moderators that may contribute to overestimation of analgesic treatment effect when testing novel compounds. Additional factors to those addressed in this review contribute to the selection of preclinical methodologies for testing analgesic efficacy, including comparability of models to human disease, the importance of avoiding unnecessary suffering in animals, cost, and available local expertise. However, preclinical researchers should pay close attention to aspects of study design, quality and potential biases in order to maximise the generalisability of their results.

      Conclusions

      Systematic review and meta-analysis can be used to investigate sources and impact of bias, and these techniques are gaining ground in animal laboratory science as shown by recent published reviews
      • Hooijmans C.R.
      • Pasker-de Jong P.C.M.
      • de Vries R.B.M.
      • Ritskes-Hoitinga M.
      The effects of long-term Omega-3 fatty acid supplementation on cognition and Alzheimer's pathology in animal models of Alzheimer's disease: a systematic review and meta-analysis.
      • Frantzias J.
      • Sena E.S.
      • Macleod M.R.
      • Salman R.A.
      Treatment of intracerebral hemorrhage in animal models: meta-analysis.
      • Hirst T.C.
      • Vesterinen H.M.
      • Sena E.S.
      • Egan K.J.
      • Macleod M.R.
      • Whittle I.R.
      Systematic review and meta-analysis of temozolomide in animal models of glioma: was clinical efficacy predicted?.
      • Lees J.S.
      • Sena E.S.
      • Egan K.J.
      • Antonic A.
      • Koblar S.A.
      • Howells D.W.
      • et al.
      Stem cell-based therapy for experimental stroke: a systematic review and meta-analysis.
      • Vesterinen H.M.
      • Sena E.S.
      • Ffrench-Constant C.
      • Williams A.
      • Chandran S.
      • Macleod M.R.
      Improving the translational hit of experimental treatments in multiple sclerosis.
      • van Drongelen J.
      • Hooijmans C.R.
      • Lotgering F.K.
      • Smits P.
      • Spaanderman M.E.A.
      Adaptive changes of mesenteric arteries in pregnancy: a meta-analysis.
      • Wever K.E.
      • Menting T.P.
      • Rovers M.
      • van der Vliet J.A.
      • Rongen G.A.
      • Masereeuw R.
      • et al.
      Ischemic preconditioning in the animal kidney, a systematic review and meta-analysis.
      and methods papers
      • Sena E.S.
      • van der Worp H.B.
      • Bath P.M.W.
      • Howells D.W.
      • Macleod M.R.
      Publication bias in reports of animal stroke studies leads to major overstatement of efficacy.
      • Hooijmans C.R.
      • de Vries R.
      • Leenaars M.
      • Curfs J.
      • Ritskes-Hoitinga M.
      Improving planning, design, reporting and scientific quality of animal experiments by using the gold standard publication checklist, in addition to the ARRIVE guidelines.
      • Hooijmans C.R.
      • Leenaars M.
      • Ritskes-Hoitinga M.
      A gold standard publication checklist to improve the quality of animal studies, to fully integrate the three Rs, and to make systematic reviews more feasible.
      • Leenaars M.
      • Hooijmans C.R.
      • van Veggel N.
      • ter Riet G.
      • Leeflang M.
      • Hooft L.
      • et al.
      A step-by-step guide to systematically identify all relevant animal studies.
      • de Vries R.B.M.
      • Hooijmans C.R.
      • Tillema A.
      • Leenaars M.
      • Ritskes-Hoitinga M.
      A search filter for increasing the retrieval of animal studies in Embase.
      . Applying such methods we have identified study-level moderators – choice of design, limited study quality and publication bias – that may result in overestimation of analgesic treatment effect. These moderators may reduce specificity for preclinical studies to predict analgesic agents that will reduce arthritis pain in man. The reviewed data suggest important publication bias in animal studies, and both researchers and editors should strive to publish negative as well as positive research findings. Particular attention to the pain modality, study quality and quantification of OA phenotype is required, both in the design of new studies, and in interpretation of results. Improvements in study design and reporting may increase the potential for preclinical research to predict effects of novel analgesic treatments in man, thereby reducing the pursuit of potentially unsuccessful clinical trials, and facilitating translation of effective novel drugs into clinical practice.

      Contributors

      AKS, DRS, PIM, VC and DAW were involved in the conception and design of the study. Data were extracted by AKS and validated by DRS, PIM and VC. AKS, DRS, VC and DAW were involved in the analysis and interpretation of data, drafting of the article and revising it critically for important intellectual content. All authors approved the final version of the manuscript.

      Funding

      None received.

      Competing interests

      DRS, VC and DAW have received support in-kind from pharmaceutical companies for testing potential analgesic drugs in animal models of OA.

      Appendix 1. Search strategy in OVID Medline

      Tabled 1
      Search terms for osteoarthritis (OA)
      • 1.
        osteoarthritis.mp. or exp Osteoarthritis, Hip/or exp Osteoarthritis/or exp Osteoarthritis, Spine/or exp Osteoarthritis, Knee/
      • 2.
        osteoarthrosis.mp.
      • 3.
        gonarthritis.mp.
      • 4.
        gonarthrosis.mp.
      • 5.
        gonitis.mp.
      • 6.
        coxarthritis.mp.
      • 7.
        coxarthrosis.mp.
      • 8.
        coxitis.mp.
      • 9.
        1 or 2 or 3 or 4 or 5 or 6 or 7 or 8
      Search terms for nociception
      • 10.
        pain.mp. or exp Pain/
      • 11.
        hyperalgesia.mp. or exp Hyperalgesia/
      • 12.
        hyperalgesic.mp.
      • 13.
        allodynia.mp.
      • 14.
        exp Pain Threshold/or exp Pain Measurement/or pain threshold*.mp.
      • 15.
        exp Nociception/
      • 16.
        nocicept*.mp.
      • 17.
        10 or 11 or 12 or 13 or 14 or 15 or 16
      Search terms for small animal models (The Species)
      • 18.
        rats.mp. or exp Rats/
      • 19.
        rat.mp.
      • 20.
        exp Mice/or mice.mp.
      • 21.
        mouse.mp.
      • 22.
        exp Guinea Pigs/or guinea pig*.mp.
      • 23.
        rabbit.mp. or exp Rabbits/
      • 24.
        rabbit*.mp.
      • 25.
        18 or 19 or 20 or 21 or 22 or 23 or 24
      Combine OA, nociception and the species
      • 26.
        9 and 17 and 25
      Search terms for spontaneous models of OA
      • 27.
        (“DBA/1” or “str/ort” or “c57bl/6” or “C57” or “BALB/c”).mp.
      • 28.
        dunkin-hartley.mp.
      • 29.
        hartley guinea pig*.mp.
      • 30.
        27 or 28 or 29
      Search terms for chemical models of OA
      • 31.
        iodoacetate*.mp.
      • 32.
        osmic acid*.mp.
      • 33.
        thiotepa*.mp.
      • 34.
        colchicine*.mp.
      • 35.
        nitrogen mustard*.mp.
      • 36.
        polyene antibiotic*.mp.
      • 37.
        (filipin* or philipine*).mp.
      • 38.
        vitamin A.mp.
      • 39.
        Steroid*.mp.
      • 40.
        corticosteroid*.mp.
      • 41.
        enzyme*.mp.
      • 42.
        collagenase*.mp.
      • 43.
        trypsin*.mp.
      • 44.
        hyaluronidase*.mp.
      • 45.
        papain*.mp.
      • 46.
        growth factor*.mp.
      • 47.
        hormone*.mp.
      • 48.
        cortisone*.mp.
      • 49.
        31 or 32 or 33 or 34 or 35 or 36 or 37 or 38 or 39 or 40 or 41 or 42 or 43 or 44 or 45 or 46 or 47 or 48
      Search terms for surgical models of OA
      • 50.
        surgically-induced.mp.
      • 51.
        surgically-evoked.mp.
      • 52.
        destabili*.mp.
      • 53.
        (Transect* adj8 ligament*).mp.
      • 54.
        (Transect* adj8 menisc*).mp.
      • 55.
        meniscectomy.mp.
      • 56.
        (cartilage adj8 scarification*).mp.
      • 57.
        (osteochondral adj8 (fragment or fracture or chip)).mp.
      • 58.
        ovariectomi*.mp.
      • 59.
        50 or 51 or 52 or 53 or 54 or 55 or 56 or 57 or 58
      Combine spontaneous or chemical or surgical models of OA
      • 60.
        30 or 49 or 59
      combine [OA, nociception and the species] and [models of OA]
      • 61.
        26 and 60

      Appendix 2. Checklist for study quality and potential risk of bias as judged by the quality of reporting. Each item was scored as 1 if reported satisfactorily, and 0 if not (unclear risk of bias). Maximum score was 8

      Tabled 1
      Risk of biasCriteriaExplanation
      Selection bias
      • 1.
        Randomisation
      Random assignment of animals to experimental (treatment) or control groups
      • 2.
        Allocation concealment
      Concealing the allocation sequence from those assigning animals to experimental and control groups, until the moment of assignment.
      Detection bias
      • 3.
        Blinding
      Keeping the persons who perform the experiment, collect data, and assess outcome unaware of the treatment allocation.
      Other sources of bias
      • 4.
        Evidence of induced OA pain prior to pharmacological study
      Timeline showing the effect of the induced OA on change in pain-related behavioural outcome measures over the course of the experiment
      • 5.
        Evidence presented to demonstrate that the animals had OA
      Histological, macroscopic, microscopic or X-ray evidence
      • 6.
        Sample size calculation before start of experiment
      • 7.
        Statement regarding potential conflict of interest
      • 8.
        Statement of compliance with animal welfare regulations

      Appendix 3. Forest plot presenting the SMD (95% CIs) between drug-and vehicle-treated rats with knee OA

      Figure thumbnail fx1

      Appendix 4. Pooled SMDs and 95% CIs for the analgesic treatment effect of NSAIDs and opioids, by Weeks 1–5 post-induction of OA. Positive increasing SMDs indicate analgesic efficacy. One study may involve the testing of more than one drug and weekly time point

      Figure thumbnail fx2
      Tabled 1
      Time after OA inductionWeek 1Week 2Week 3Week 4Week 5
      NSAIDs
       SMD1.251.061.211.200.84
       95% CIs0.73, 1.770.42, 1.690.79, 1.630.62, 1.790.20, 1.49
       No. of studies7612114
       No. of comparisons171121196
       Total sample size25421232631076
      Opioids
       SMD1.941.471.911.743.07
       95% CIs1.45, 2.420.83, 2.111.40, 2.410.51, 2.970.02, 6.12
       No. of studies52522
       No. of comparisons731323
       Total sample size107612093637

      Appendix 5. Forest plot of comparisons measuring the analgesic treatment effect of NSAIDs and opioids between drug-treated and vehicle-treated rats with knee OA, by pain modality. A summary table is presented in Fig. 2. SMD = standardised mean difference, 95% CIs = 95% confidence intervals

      Figure thumbnail fx3
      Figure thumbnail fx4
      Figure thumbnail fx5
      Figure thumbnail fx6

      Appendix 6. The impact of study-level moderators on the magnitude of the analgesic treatment effect (n = 102 unless stated otherwise). All variables (apart from NSAID vs opioid) have been adjusted for the class of drug

      Tabled 1
      Individual variables/groups of dummiesBZP
      Class of drug
      NSAID vs opioid−0.80−3.250.001
      Values in bold are statistically significant at P < 0.05.
      Pain modality (n = 100)
      Thermally-evoked pain with only two observations from one study was excluded from the analysis, n = 100.
      Weight bearing vs grip strength−2.66−4.60< 0.001
      Values in bold are statistically significant at P < 0.05.
      Mechanically-evoked pain vs grip strength−2.29−3.83< 0.001
      Values in bold are statistically significant at P < 0.05.
      Movement-evoked pain vs grip strength−3.22−5.23< 0.001
      Values in bold are statistically significant at P < 0.05.
      Movement-evoked pain vs mechanically-evoked pain−0.92−2.780.006
      Values in bold are statistically significant at P < 0.05.
      Weight bearing vs mechanically-evoked pain−0.37−1.410.157
      Weight-bearing vs movement-evoked pain0.561.820.068
      Timing of the outcome measurement (post-induction of OA): Weeks 15
      Week 2 vs Week 1−0.29−0.790.429
      Week 3 vs Week 1−0.06−0.200.838
      Week 4 vs Week 1−0.12−0.360.716
      Week 5 vs Week 1−0.13−0.300.763
      Week 3 vs Week 20.220.660.511
      Week 4 vs Week 20.170.450.654
      Week 5 vs Week 20.150.320.748
      Week 4 vs Week 3−0.06−0.190.848
      Week 5 vs Week 3−0.07−0.170.866
      Week 5 vs Week 4−0.01−0.030.978
      Method of induction
      MIA vs surgical induction−0.42−1.030.302
      Dose of MIA (n = 94)
      Observations that used a surgical method of OA induction were excluded, n = 94.
      MIA 2 mg vs MIA 1 mg0.651.900.057
      MIA 3 mg vs MIA 1 mg0.070.250.801
      MIA 3 mg vs MIA 2 mg−0.58−1.880.060
      Rat strain
      Wistar vs Sprague Dawley−0.15−0.650.513
      Lewis vs Sprague Dawley0.290.620.536
      Lewis vs Wistar0.440.910.362
      Sample size per comparison
      Sample size ≤14 vs >14−0.07−0.330.742
      Route of administration
      Oral vs intraperitoneal−2.55−3.94<0.001
      Values in bold are statistically significant at P < 0.05.
      Subcutaneous vs intraperitoneal−2.36−3.44<0.001
      Values in bold are statistically significant at P < 0.05.
      Oral vs subcutaneous−0.19−0.680.499
      Administration of the drug
      Acute vs chronic0.150.650.515
      Overall study quality
      Total quality score−0.34−3.64<0.001
      Values in bold are statistically significant at P < 0.05.
      Study quality sub-components
      None of the studies reported allocation concealment or sample size calculation, and all gave a statement of compliance with animal welfare regulations. These categories were excluded from the analysis.
      Randomisation to groups−0.50−1.790.074
      Blinding of outcome assessment−0.37−1.450.148
      Evidence of induced OA pain prior to pharmacological study−0.16−0.550.580
      Evidence presented to demonstrate that the animals had OA−0.61−2.450.014
      Values in bold are statistically significant at P < 0.05.
      Conflict of interest statement0.070.160.872
      Excluded variables: Sex (all animals were male); age and weight (insufficient reporting).
      Values in bold are statistically significant at P < 0.05.
      Thermally-evoked pain with only two observations from one study was excluded from the analysis, n = 100.
      Observations that used a surgical method of OA induction were excluded, n = 94.
      § None of the studies reported allocation concealment or sample size calculation, and all gave a statement of compliance with animal welfare regulations. These categories were excluded from the analysis.

      References

        • Gwilym S.E.
        • Pollard T.C.
        • Carr A.J.
        Understanding pain in osteoarthritis.
        J Bone Joint Surg Br. 2008; 90: 280-287
        • Mease P.J.
        • Hanna S.
        • Frakes E.P.
        • Altman R.D.
        Pain mechanisms in osteoarthritis: understanding the role of central pain and current approaches to its treatment.
        J Rheumatol. 2011; 38: 1546-1551
        • Sofat N.
        • Ejindu V.
        • Kiely P.
        What makes osteoarthritis painful? The evidence for local and central pain processing.
        Rheumatology. 2011; 50: 2157-2165
        • Schnitzer T.J.
        Update on guidelines for the treatment of chronic musculoskeletal pain.
        Clin Rheumatol. 2006; 25: S22-S29
        • Teeple E.
        • Jay G.D.
        • Elsaid K.A.
        • Fleming B.C.
        Animal models of osteoarthritis: challenges of model selection and analysis.
        AAPS J. 2013; 15: 438-446
        • Gregory M.H.
        • Capito N.
        • Kuroki K.
        • Stoker A.M.
        • Cook J.L.
        • Sherman S.L.
        A review of translational animal models for knee osteoarthritis.
        Arthritis. 2012; 2012 (Epub 762012 Dec 764627): 764621https://doi.org/10.1155/2012/764621
        • Dworkin R.H.
        • Turk D.C.
        • Basch E.
        • Berger A.
        • Cleeland C.
        • Farrar J.T.
        • et al.
        Considerations for extrapolating evidence of acute and chronic pain analgesic efficacy.
        Pain. 2011; 152: 1705-1708
        • Dworkin R.H.
        • Turk D.C.
        • Katz N.P.
        • Rowbotham M.C.
        • Peirce-Sandner S.
        • Cerny I.
        • et al.
        Evidence-based clinical trial design for chronic pain pharmacotherapy: a blueprint for ACTION.
        Pain. 2011; 152: S107-S115
        • Perel P.
        • Roberts I.
        • Sena E.
        • Wheble P.
        • Briscoe C.
        • Sandercock P.
        • et al.
        Comparison of treatment effects between animal experiments and clinical trials: systematic review.
        Br Med J. 2007; 334: 197-200
        • Sena E.
        • van der Worp H.B.
        • Howells D.
        • Macleod M.
        How can we improve the pre-clinical development of drugs for stroke?.
        Trends Neurosci. 2007; 30: 433-439
        • Sena E.S.
        • van der Worp H.B.
        • Bath P.M.W.
        • Howells D.W.
        • Macleod M.R.
        Publication bias in reports of animal stroke studies leads to major overstatement of efficacy.
        PloS Biol. 2010; 8
        • NICE
        The Care and Management of Osteoarthritis in Adults.
        National Institute for Health and Clinical Excellence, London2008
        • Vincent T.L.
        • Williams R.O.
        • Maciewicz R.
        • Silman A.
        • Garside P.
        Mapping pathogenesis of arthritis through small animal models.
        Rheumatology. 2012; 51: 1931-1941
        • Boyce-Rustay J.M.
        • Simler G.H.
        • McGaraughty S.
        • Chu K.L.
        • Wensink E.J.
        • Vasudevan A.
        • et al.
        Characterization of fasudil in preclinical models of pain.
        J Pain. 2010; 11: 941-949
        • Kuduk S.D.
        • Di Marco C.N.
        • Bodmer-Narkevitch V.
        • Cook S.P.
        • Cato M.J.
        • Jovanovska A.
        • et al.
        Synthesis, structure-activity relationship, and pharmacological profile of analogs of the ASIC-3 inhibitor A-317567.
        ACS Chem Neurosci. 2010; 1: 19-24
        • Pulichino A.M.
        • Rowland S.
        • Wu T.
        • Clark P.
        • Xu D.
        • Mathieu M.C.
        • et al.
        Prostacyclin antagonism reduces pain and inflammation in rodent models of hyperalgesia and chronic arthritis.
        J Pharmacol Exp Ther. 2006; 319: 1043-1050
        • Pomonis J.D.
        • Boulet J.M.
        • Gottshall S.L.
        • Phillips S.
        • Sellers R.
        • Bunton T.
        • et al.
        Development and pharmacological characterization of a rat model of osteoarthritis pain.
        Pain. 2005; 114: 339-346
        • Hooijmans C.R.
        • de Vries R.B.M.
        • Rovers M.M.
        • Gooszen H.G.
        • Ritskes-Hoitinga M.
        The effects of probiotic supplementation on experimental acute pancreatitis: a systematic review and meta-analysis.
        PLoS One. 2012; 7
        • Hooijmans C.R.
        • Pasker-de Jong P.C.M.
        • de Vries R.B.M.
        • Ritskes-Hoitinga M.
        The effects of long-term Omega-3 fatty acid supplementation on cognition and Alzheimer's pathology in animal models of Alzheimer's disease: a systematic review and meta-analysis.
        J Alzheimers Dis. 2012; 28: 191-209
        • Borenstein M.
        • Hedges L.
        • Higgins J.
        • Rothstein H.
        Introduction to Meta Analysis.
        John Wiley & Sons, Ltd, Chichester2010
        • Wilson D.
        Meta-analysis.
        ([Accessed on 5 Jan 2014])
        • Vonsy J.L.
        • Ghandehari J.
        • Dickenson A.H.
        Differential analgesic effects of morphine and gabapentin on behavioural measures of pain and disability in a model of osteoarthritis pain in rats.
        Eur J Pain. 2009; 13: 786-793
        • Boersen N.
        • Lee T.
        • Hui H.
        Development of preclinical formulations for toxicology studies.
        in: Faqi A. A Comprehensive Guide to Toxicology in Preclinical Drug Development. Academic Press, San Diego, CA2013
        • Hedenqvist P.
        • Hellebrekers L.J.
        Laboratory animal analgesia, anesthesia, and euthanasia.
        in: Hau J. Van Hoosier G.L. Handbook of Laboratory Animal Science. CRC Press, Boca Raton, Florida2003: 413-455
        • Turner P.V.
        • Brabb T.
        • Pekow C.
        • Vasbinder M.A.
        Administration of substances to laboratory animals: routes of administration and factors to consider.
        J Am Assoc Lab Anim Sci. 2011; 50: 600-613
        • Kehl L.J.
        • Trempe T.M.
        • Hargreaves K.M.
        A new animal model for assessing mechanisms and management of muscle hyperalgesia.
        Pain. 2000; 85: 333-343
        • Wacnik P.W.
        • Kehl L.J.
        • Trempe T.M.
        • Ramnaraine M.L.
        • Beitz A.J.
        • Wilcox G.L.
        Tumor implantation in mouse humerus evokes movement-related hyperalgesia exceeding that evoked by intramuscular carrageenan.
        Pain. 2003; 101: 175-186
        • Chandran P.
        • Pai M.
        • Blomme E.A.
        • Hsieh G.C.
        • Decker M.W.
        • Honore P.
        Pharmacological modulation of movement-evoked pain in a rat model of osteoarthritis.
        Eur J Pharmacol. 2009; 613: 39-45
        • Kilkenny C.
        • Browne W.J.
        • Cuthill I.C.
        • Emerson M.
        • Altman D.G.
        Improving bioscience research reporting: the ARRIVE guidelines for reporting animal research.
        Osteoarthritis & Cartilage. 2012; 20: 256-260
        • Frantzias J.
        • Sena E.S.
        • Macleod M.R.
        • Salman R.A.
        Treatment of intracerebral hemorrhage in animal models: meta-analysis.
        Ann Neurol. 2011; 69: 389-399
        • Hirst T.C.
        • Vesterinen H.M.
        • Sena E.S.
        • Egan K.J.
        • Macleod M.R.
        • Whittle I.R.
        Systematic review and meta-analysis of temozolomide in animal models of glioma: was clinical efficacy predicted?.
        Br J Cancer. 2013; 108: 64-71
        • Lees J.S.
        • Sena E.S.
        • Egan K.J.
        • Antonic A.
        • Koblar S.A.
        • Howells D.W.
        • et al.
        Stem cell-based therapy for experimental stroke: a systematic review and meta-analysis.
        Int J Stroke. 2012; 7: 582-588
        • Vesterinen H.M.
        • Sena E.S.
        • Ffrench-Constant C.
        • Williams A.
        • Chandran S.
        • Macleod M.R.
        Improving the translational hit of experimental treatments in multiple sclerosis.
        Mult Scler J. 2010; 16: 1044-1055
        • van Drongelen J.
        • Hooijmans C.R.
        • Lotgering F.K.
        • Smits P.
        • Spaanderman M.E.A.
        Adaptive changes of mesenteric arteries in pregnancy: a meta-analysis.
        Am J Physiol Heart Circ Physiol. 2012; 303: H639-H657
        • Wever K.E.
        • Menting T.P.
        • Rovers M.
        • van der Vliet J.A.
        • Rongen G.A.
        • Masereeuw R.
        • et al.
        Ischemic preconditioning in the animal kidney, a systematic review and meta-analysis.
        PLoS One. 2012; 7
        • Hooijmans C.R.
        • de Vries R.
        • Leenaars M.
        • Curfs J.
        • Ritskes-Hoitinga M.
        Improving planning, design, reporting and scientific quality of animal experiments by using the gold standard publication checklist, in addition to the ARRIVE guidelines.
        Br J Pharmacol. 2011; 162: 1259-1260
        • Hooijmans C.R.
        • Leenaars M.
        • Ritskes-Hoitinga M.
        A gold standard publication checklist to improve the quality of animal studies, to fully integrate the three Rs, and to make systematic reviews more feasible.
        Atla-Altern Lab Anim. 2010; 38: 167-182
        • Leenaars M.
        • Hooijmans C.R.
        • van Veggel N.
        • ter Riet G.
        • Leeflang M.
        • Hooft L.
        • et al.
        A step-by-step guide to systematically identify all relevant animal studies.
        Lab Anim. 2012; 46: 24-31
        • de Vries R.B.M.
        • Hooijmans C.R.
        • Tillema A.
        • Leenaars M.
        • Ritskes-Hoitinga M.
        A search filter for increasing the retrieval of animal studies in Embase.
        Lab Anim. 2011; 45: 268-270
        • Ahn K.
        • Smith S.E.
        • Liimatta M.B.
        • Beidler D.
        • Sadagopan N.
        • Dudley D.T.
        • et al.
        Mechanistic and pharmacological characterization of PF-04457845: a highly potent and selective fatty acid amide hydrolase inhibitor that reduces inflammatory and noninflammatory pain.
        J Pharmacol Exp Ther. 2011; 338: 114-124
        • Ashraf S.
        • Mapp P.I.
        • Walsh D.A.
        Contributions of angiogenesis to inflammation, joint damage, and pain in a rat model of osteoarthritis.
        Arthritis Rheum. 2011; 63: 2700-2710
        • Beyreuther B.
        • Callizot N.
        • Stohr T.
        Antinociceptive efficacy of lacosamide in the monosodium iodoacetate rat model for osteoarthritis pain.
        Arthritis Res Ther. 2007; 9: R14
        • Bley K.R.
        • Bhattacharya A.
        • Daniels D.V.
        • Gever J.
        • Jahangir A.
        • O'Yang C.
        • et al.
        RO1138452 and RO3244794: characterization of structurally distinct, potent and selective IP (prostacyclin) receptor antagonists.
        Br J Pharmacol. 2006; 147: 335-345
        • Bove S.E.
        • Calcaterra S.L.
        • Brooker R.M.
        • Huber C.M.
        • Guzman R.E.
        • Juneau P.L.
        • et al.
        Weight bearing as a measure of disease progression and efficacy of anti-inflammatory compounds in a model of monosodium iodoacetate-induced osteoarthritis.
        Osteoarthritis and Cartilage. 2003; 11: 821-830
        • Bove S.E.
        • Laemont K.D.
        • Brooker R.M.
        • Osborn M.N.
        • Sanchez B.M.
        • Guzman R.E.
        • et al.
        Surgically induced osteoarthritis in the rat results in the development of both osteoarthritis-like joint pain and secondary hyperalgesia.
        Osteoarthritis & Cartilage. 2006; 14: 1041-1048
        • Castro R.R.
        • Cunha F.Q.
        • Silva F.S.
        • Rocha F.A.C.
        A quantitative approach to measure joint pain in experimental osteoarthritis – evidence of a role for nitric oxide.
        Osteoarthritis and Cartilage. 2006; 14: 769-776
        • Combe R.
        • Bramwell S.
        • Field M.J.
        The monosodium iodoacetate model of osteoarthritis: a model of chronic nociceptive pain in rats?.
        Neurosci Lett. 2004; 370: 236-240
        • Cowart M.
        • Hsieh G.
        • Black L.A.
        • Zhan C.
        • Gomez E.J.
        • Pai M.
        • et al.
        Pharmacological characterization of A-960656, a histamine H(3) receptor antagonist with efficacy in animal models of osteoarthritis and neuropathic pain.
        Eur J Pharmacol. 2012; 684: 87-94
        • Ferland C.E.
        • Laverty S.
        • Beaudry F.
        • Vachon P.
        Gait analysis and pain response of two rodent models of osteoarthritis.
        Pharmacol Biochem Behav. 2011; 97: 603-610
        • Fernihough J.
        • Gentry C.
        • Malcangio M.
        • Fox A.
        • Rediske J.
        • Pellas T.
        • et al.
        Pain related behaviour in two models of osteoarthritis in the rat knee.
        Pain. 2004; 112: 83-93
        • Ferreira-Gomes J.
        • Adaes S.
        • Mendonca M.
        • Castro-Lopes J.M.
        Analgesic effects of lidocaine, morphine and diclofenac on movement-induced nociception, as assessed by the Knee-Bend and CatWalk tests in a rat model of osteoarthritis.
        Pharmacol Biochem Behav. 2012; 101: 617-624
        • Fitzpatrick L.R.
        • Green C.
        • Maines L.W.
        • Smith C.D.
        Experimental osteoarthritis in rats is attenuated by ABC294640, a selective inhibitor of sphingosine kinase-2.
        Pharmacology. 2011; 87: 135-143
        • Hsieh G.C.
        • Chandran P.
        • Salyers A.K.
        • Pai M.
        • Zhu C.Z.
        • Wensink E.J.
        • et al.
        H4 receptor antagonism exhibits anti-nociceptive effects in inflammatory and neuropathic pain models in rats.
        Pharmacol Biochem Behav. 2010; 95: 41-50
        • Hsieh G.C.
        • Honore P.
        • Pai M.
        • Wensink E.J.
        • Chandran P.
        • Salyers A.K.
        • et al.
        Antinociceptive effects of histamine H3 receptor antagonist in the preclinical models of pain in rats and the involvement of central noradrenergic systems.
        Brain Res. 2010; 1354: 74-84
        • Ivanavicius S.P.
        • Ball A.D.
        • Heapy C.G.
        • Westwood F.R.
        • Murray F.
        • Read S.J.
        Structural pathology in a rodent model of osteoarthritis is associated with neuropathic pain: increased expression of ATF-3 and pharmacological characterisation.
        Pain. 2007; 128: 272-282
        • Kang M.
        • Jung I.
        • Hur J.
        • Kim S.H.
        • Lee J.H.
        • Kang J.Y.
        • et al.
        The analgesic and anti-inflammatory effect of WIN-34B, a new herbal formula for osteoarthritis composed of Lonicera japonica Thunb and Anemarrhena asphodeloides BUNGE in vivo.
        J Ethnopharmacol. 2010; 131: 485-496
        • Okun A.
        • Liu P.
        • Davis P.
        • Ren J.
        • Remeniuk B.
        • Brion T.
        • et al.
        Afferent drive elicits ongoing pain in a model of advanced osteoarthritis.
        Pain. 2012; 153: 924-933
        • Plummer M.S.
        • Cornicelli J.
        • Roark H.
        • Skalitzky D.J.
        • Stankovic C.J.
        • Bove S.
        • et al.
        Discovery of potent selective bioavailable phosphodiesterase 2 (PDE2) inhibitors active in an osteoarthritis pain model. Part II: optimization studies and demonstration of in vivo efficacy.
        Bioorganic Med Chem Lett. 2013; 23: 3443-3447
        • Rashid M.H.
        • Theberge Y.
        • Elmes S.J.
        • Perkins M.N.
        • McIntosh F.
        Pharmacological validation of early and late phase of rat mono-iodoacetate model using the Tekscan system.
        Eur J Pain. 2013; 17: 210-222
        • Yao B.B.
        • Hsieh G.C.
        • Frost J.M.
        • Fan Y.
        • Garrison T.R.
        • Daza A.V.
        • et al.
        In vitro and in vivo characterization of A-796260: a selective cannabinoid CB2 receptor agonist exhibiting analgesic activity in rodent pain models.
        Br J Pharmacol. 2008; 153: 390-401
        • Yoshimi E.
        • Kumakura F.
        • Hatori C.
        • Hamachi E.
        • Iwashita A.
        • Ishii N.
        • et al.
        Antinociceptive effects of AS1892802, a novel Rho kinase inhibitor, in rat models of inflammatory and noninflammatory arthritis.
        J Pharmacol Exp Ther. 2010; 334: 955-963